Key Features
Democratising access to high-level game analysis — no coach required.
Match History Analysis
Instantly processes the last games to surface trends, patterns, and recurring mistakes invisible in single-game reviews.
Deep KPI Insights
Goes beyond basic KDA — tracks CS/min, Gold/min, Vision Score, Damage Share, Kill Participation, and Objective Control.
Personalised AI Coaching
Gemini adapts its critique to champion pool and role played, covering mechanics, macro-play, and strategic recommendations.
Data Export Utility
Standalone export.py dumps match data into CSV, JSON, and TOON formats for offline analysis or dataset creation.
Tech Stack
Modern Python ecosystem built for async performance, AI integration, and scalability.
How It Works
A modular pipeline from Discord slash command to AI coaching report — in one flow.
User Interaction
A player invokes the /coach slash command with their Riot ID and Tagline directly inside Discord.
Data Ingestion
Resolves PUUID via the Account API, fetches the last match IDs, and retrieves detailed participant timelines for each game.
Data Processing
Filters and normalises player metrics into structured Pandas DataFrames — KDA, CS/min, Vision, Gold efficiency, and more.
AI Analysis
A context-aware prompt sends aggregated match data to Gemini, which returns a structured critique of mechanics, macro, and strategy.
Response Delivery
The coaching report is chunked to respect Discord's message limits and delivered back to the player in the channel.
Architecture Overview
Designed for modularity — each component can be extended or replaced independently.
Discord Layer
- • Asynchronous slash commands
- • Message chunking (2 000-char limit)
- • Error handling & user feedback
Data Pipeline
- • MatchV5, SummonerV4, AccountV1
- • Pandas DataFrames & aggregation
- • TOON serialisation for LLM context
AI Layer
- • Gemini 2.5 Flash
- • Context-aware prompt engineering
- • Structured critique output
Development Roadmap
From a single slash command to a full coaching platform.
Core Coaching Bot
Slash command, Riot API integration, Gemini-powered coaching report, Discord message chunking.
Frame-by-Frame Timeline Analysis
Fetch granular timeline data to analyse specific skirmishes and teamfight positioning at a per-minute level.
Visual Reports
Generate Gold/XP lead graphs with matplotlib and embed them directly in Discord responses.
Player Profiles & History
Store coaching sessions in a database to track improvement over time and compare across patches.
Multi-Region Support
Enhanced routing logic to dynamically handle all Riot API regions without manual configuration.